Prognostic value of lymphovascular invasion in stage II colorectal cancer patients with an inadequate examination of lymph nodes

Background Lymphovascular invasion (LVI) is defined as the presence of cancer cells in lymphatics or blood vessels. This study aimed to evaluate the prognostic value of LVI in stage II colorectal cancer (CRC) patients with inadequate examination of lymph nodes (ELNs) and further combined LVI with the TNM staging system to determine the predictive efficacy for CRC prognosis. Adjuvant chemotherapy (ACT) was then evaluated for stage II CRC patients with LVI positivity (LVI+). Methods In order to avoid the effects of different ACT regimens, among 409 stage II patients, we chose 121 patients who received FOLFOX regimen and the 144 patients who did not receive ACT as the object of study. LVI was examined by hematoxylin-eosin (HE) staining. Kaplan-Meier analysis followed by a log-rank test was used to analyze survival rates. Univariate and multivariate analyses were performed using a Cox proportional hazards model. Harrell’s concordance index (C-index) was used to evaluate the accuracy of different systems in predicting prognosis. Results The LVI+ status was significantly associated with pT stage, degree of differentiation, tumor stage, serum CEA and CA19-9 levels, perineural invasion (PNI), tumor budding (TB), and KRAS status. The 5-year overall survival (OS) rate of stage II patients with < 12 ELNs and LVI+ was less than stage IIIA. Multivariate analyses showed that LVI, pT-stage, serum CEA and CA19-9 levels, PNI, TB, and KRAS status were significant prognostic factors for stage II patients with < 12 ELNs. The 8th TNM staging system combined with LVI showed a higher C-index than the 8th TNM staging system alone (C-index, 0.895 vs. 0.833). Among patients with LVI+, the ACT group had a significantly higher 5-year OS and 5-year disease-free survival (DFS) than the surgery alone (SA) group (5-year OS, 66.7% vs. 40.9%, P = 0.004; 5-year DFS, 64.1% vs. 36.3%, P = 0.002). Conclusions LVI is an independent prognostic risk factor for stage II CRC patients. Combining LVI with the 8th TNM staging system improved the predictive accuracy for CRC prognosis. ACT in stage II CRC patients with LVI+ is beneficial for survival.


Introduction
Colorectal cancer (CRC) is the third most common malignancy and the fourth leading cause of tumorrelated deaths worldwide [1]. Although advances have been achieved in early detection and effective treatment, the survival rate of CRC is still poor [2]. Among all CRC patients, approximately one-third are diagnosed as stage II [3]. The National Comprehensive Cancer Network (NCCN) guidelines recommend adjuvant chemotherapy (ACT) for stage III and IV CRC [4]. For stage II CRC, the current guidelines recommend that ACT should be considered for patients at high risk for recurrence [5].
In addition, the current guidelines recommend that at least 12 lymph nodes (LNs) should be examined for nodal evolution [6]. Adequate LN retrieval from the specimen is essential to ensure accuracy in nodal staging [7]. An inadequate examination of lymph nodes (ELNs) may cause a false-negative result or a lower pN stage [8].
Lymphovascular invasion (LVI) is defined as the presence of cancer cells in lymphatics or blood vessels and is considered to be an early step in lymph node metastasis ( Fig. 1) [1]. Many studies have reported that LVI positivity (LVI+) is a critical prognostic indicator in some cancers, including breast, bladder, and gastric cancers [9][10][11]. It has been reported that the presence of LVI in CRC varies from 4.1 to 89.5% [12]. Currently, few studies have focused on LVI in stage II CRC with inadequate ELNs. Moreover, no study involving the combination of LVI and the TNM staging system in CRC patients has been published.
Tumor budding (TB) was defined as a single tumor cell or cluster comprising less than five cells at the invasive front [13]. At the International Tumor Budding Consensus Conference (ITBCC), it was clearly stated that tumor budding is an independent prognostic factor for CRC. The reason why tumor budding has not yet developed into routine clinical practice is because there is no consensus on the scoring method. This study used the methods proposed by ITBCC in clinical practice and studied the relationship between TB and the survival of patients with stage II CRC with ELNs. Cutoffs used were the following: low, 0-9 buds; and high, ≥ 10 buds (Fig.  2a) [13]. Perineural invasion (PNI) is the process of nerve tumor infiltration, including tumor cells located in the three layers of the peripheral nerve sheath or adjacent to the nerve, and involving at least one third of its surroundings [14,15]. PNI has become a key pathological feature of many malignant tumors, including malignant tumors of the stomach, colon and rectum, pancreas, and biliary tract (Fig. 2b) [16][17][18][19]. At present, there is no consensus regarding the inclusion of PNI in staging, although PNI has been proven to be a sign of poor survival in colorectal cancer.
We conducted the current study to evaluate the prognostic value of LVI in stage II CRC patients with inadequate ELNs and combined LVI with the tumor-node-metastasis (TNM) staging system to determine the predictive efficacy for CRC prognosis. ACT was then evaluated for stage II CRC patients with LVI+.

Patients
We retrospectively examined the clinicopathologic records of CRC patients who were treated at the Third Affiliated Hospital of Soochow University between February 2007 and February 2013. The inclusion criteria were as follows: adenocarcinoma confirmed by histopathology, curative resection with lymphadenectomy, no neoadjuvant chemoradiation, complete clinicopathologic records, absence of distant metastases. The exclusion criteria were as follows: received neoadjuvant chemoradiation, incomplete clinicopathologic records, lost to follow-up, distant metastases. CRC stage was classified according to the 8th edition of the American Joint Committee on Cancer (AJCC) TNM staging system. Follow-up was carried out by telephone calls, emails, and on-site visits. Informed written consent was obtained from all CRC patients. The study was approved by the Ethics Committee of the Third Affiliated Hospital of Soochow University.

ACT regimens for stage II CRC patients
After curative resection, some stage II CRC patients chose to receive ACT for further treatment. The ACT regimen was established by our clinicians based on the patient's general performance, clinicopathologic features, and operative factors.

Data collection and LVI examination
Patient medical records were reviewed to obtain clinicopathologic data. Age, sex, tumor size, tumor location, LVI, TNM stage, degree of differentiation, ELNs, serum CEA and CA19-9 levels, perineural invasion, tumor budding, and KRAS status were recorded. Specimens were fixed in formalin, then cut into multiple slices. The histopathological examination was performed using a 5-mm-thick longitudinal whole tissue section. Slices were then embedded in paraffin and stained with hematoxylin-eosin (HE). All H&E slides were evaluated by at least two experienced pathologists, who independently assessed LVI, TB, and PNI. Lymphovascular invasion (LVI) is defined as the presence of cancer cells in lymphatics or blood vessels [1]. Tumor budding (TB) was defined as a single tumor cell or cluster comprising less than five cells at the invasive front [13]. Perineural invasion (PNI) is the process of nerve tumor infiltration, including tumor cells located in the three layers of the peripheral nerve sheath or adjacent to the nerve, and involving at least one third of its surroundings [14,15]. The Amplified Refractory Mutation System (ARMS) was used to detect KRAS mutations in colorectal cancer.

Statistical analysis
All analyses were performed using the SPSS (version 25.0 software; IBM, Chicago, IL, USA) and R software (version 3.0.0; www.r-project.org). Statistical significance was tested using Student's t-test and chisquared test. Univariate and multivariate analyses were performed using a Cox proportional hazards model. Kaplan-Meier analysis followed by a log-rank test was used to analyze survival rates. Harrell's concordance index (C-index) was used to evaluate the accuracy of different systems in predicting prognosis. All statistical analyses were two-sided, and a P < 0.05 was considered statistically significant.

Patient characteristics
One thousand four hundred and twenty CRC patients from February 2007 to February 2013 who met the inclusion criteria were evaluated in this study. The clinicopathologic characteristics of 1420   patients who received FOLFOX regimen and the 144 patients who did not receive ACT as the object of study. Finally, these 265 stage II patients were analyzed in this study whose clinicopathologic characteristics are listed in Table 1. One hundred patients (37.7%) were LVI+ and 165 (62.3%) were LVI−. We divided these stage II patients into ACT and surgery alone (SA) groups.

Occurrence of LVI in stage II CRC patients
As shown in Table 2, the incidence of LVI in 265 stage II CRC patients is listed based on clinicopathologic characteristics. The LVI status was significantly associated with pT stage, degree of differentiation, tumor stage, serum CEA and CA19-9 levels, perineural invasion, tumor budding, and KRAS status. The incidence of LVI in pT3, pT4a, and pT4b stages was 22.9%, 47.7%, and 64.3%, respectively. There was a statistically significant difference between LVI and pT stage. No significant difference existed with respect to sex, age, tumor size, tumor site, and ELNs.

Univariate and multivariate analyses for the prognosis of stage II patients with < 12 ELNs
Owing to the specific characteristics of stage II patients with < 12 ELNs, the prognostic factors were further analyzed. Univariate analyses showed that LVI, pT-stage, degree of differentiation, and CEA and CA19-9 levels were significant prognostic factors for stage II patients with < 12 ELNs; further multivariate analysis identified that LVI, pT-stage, degree of differentiation, CEA and CA19-9 levels, PNI, TB, and KRAS status were significant prognostic factors for stage II patients with < 12 ELNs (all P < 0.05) ( Table 3).

Improvement of the 8th TNM staging system
Because of the similarity in 5-year OS rates between stage II LVI+ patients with < 12 ELNs and stages IIIA and IIIB patients, we combined LVI with the 8th TNM staging system. A comparison was made to estimate the prognostic value between the new system and the 8th TNM staging system (Table 4). Stage II LVI+ patients with < 12 ELNs were upgraded to stage III, while stage II LVI− patients with < 12 ELNs remained stage II. The 8th TNM staging system combined with LVI had a higher C-index than the 8th TNM staging system alone (C-index, 0.895 vs. 0.833), which indicates a better prognostic value for CRC patients.

Relationship between LVI and ACT in stage II CRC patients
In addition to analyzing OS, we also analyzed diseasefree survival (DFS), especially in stage II CRC patients. The 5-year DFS rate of the LVI− group was greater than the LVI+ group (78.2% vs. 58.0%; P < 0.001; Fig. 4b). We  Fig. 7d).

Discussion
CRC has become a major public health issue worldwide, with 1.4 million new cases and 0.7 million deaths each year [20]. Curative surgery with or without chemotherapy and radiotherapy is the mainstay of treatment for CRC [21]. ACT is recommended for stages III and IV CRC patients [4]. It has been reported that ACT improves OS in stage III patients [22]; however, the benefit of ACT in stage II CRC patients is controversial. The 8th TNM staging system remains the most important prognostic indicator for CRC patients [8]. For pN stage patients, at least 12 ELNs are recommended to avoid false-negative prognostication; however, it is unavoidable that some cases have < 12 ELNs, which may interfere with the nodal classification and even influence prognosis. LVI has been reported to occur in 10-89.5% of CRC patients [23], which is also considered to increase the risk for micrometastases in localized cancer [24]. Thus, in this study, we focused on stage II LVI+ patients with < 12 ELNs.
As a common histopathologic finding, LVI serves as a prognostic risk factor in many carcinomas [23,25,26]. In this study, the LVI+ rate was 34.2% among all CRC patients and 37.7% of stage II CRC patients, which is in agreement with previous studies [20]. Differences in the LVI+ rate might reflect the diagnostic technique used and the number of patients in various studies [27]. In our 100 stage II LVI+ patients, LVI was significantly correlated with pT stage, degree of differentiation, tumor stage, serum CEA and CA19-9 levels, KRAS status, TB, and PNI. Similar to our results, Lim et al. [23] reported an association between LVI and more advanced T and N categories, higher pre-CEA levels, and worse tumor grade. Al-Sukhni et al. [28] also concluded that LVI is related to several factors in patients with advanced CRC, including larger size, more advanced T stage, LN involvement, and distant metastasis. Zhong et al. [1] also showed that LVI is significantly associated with an increased CEA level, increased tumor differentiation, and advanced tumor stage. These studies all support our results. Thus, it has been suggested that the presence of LVI should serve as an indicator of extending the resection area [29].
Survival analyses were conducted in this study. The 5-year OS and DFS rates in stage II LVI+ patients were significantly less than LVI− patients. We even found that stage II LVI+ patients with < 12 ELNs had  Table 3 Univariate and multivariate analysis of prognostic factors for stage II patients with <12 ELNs. We analyzed the prognostic factors for those patients. As listed in Table 3, univariate analysis showed that LVI, pT-stage, degree of differentiation, serum CEA and CA19-9 levels, perineural invasion, tumor budding, and KRAS status were significant prognostic factors for those patients KRAS status, TB, and PNI were significant prognostic factors for stage II CRC patients. Similar to our conclusion, it has been shown that LVI is an independent poor prognostic factor for survival among CRC patients [30]. Huh et al. [31] also reported that N0 stage CRC patients, especially stage II, may benefit most from the presence of LVI because these patients may have a superior response to ACT. The current metaanalysis shows that mutations in the KRAS gene appear to be associated with OS in CRC patients [32]. However, another study found that KRAS and BRAF mutations are independent poor prognostic factors for the OS of stage IV tumors rather than stages I-III tumors [33]. Jang et al. [34] concluded that KRAS mutations are significantly associated with high-grade TB; furthermore, tumors with KRAS mutations in exons 3 and 4 tended to have LVI and PNI. Marx et al. [35] concluded that higher TB status is related to higher tumor grade and stage, positive lymph nodes, and LVI; similar to our conclusion, it has been shown that TB is an independent poor prognostic factor for survival among CRC patients. Al-Sukhni et al. [28] reported an association between LVI, PNI, and advanced CRC and found that PNI is an independent poor prognostic marker for survival in CRC. Skancke et al. [4] also showed that LVI and PNI have an adverse effect on the survival of patients with stage II colon cancer. When LVI and PNI are present, ACT may have a protective effect.
To explore the benefit from ACT, we focused on the survival of ACT patients with or without LVI. Our results showed that ACT improved the 5-year OS and DFS rates in LVI+ patients. Several studies have reported that ACT is beneficial for stage II CRC patients [36,37]. Similar conclusions were reported by Skancke et al. [4], who demonstrated that CRC patients with high-risk factors, including LVI, can benefit from ACT. Arakawa et al. [38] reported that a significant prognostic benefit is achieved after ACT for stage IIb/c CRC patients. Lin et al. [39] enrolled 1039 stage II CRC patients and concluded that ACT improves the DFS rate in high-risk stage II CRC patients.
It has been reported that the improvement in OS and DFS rates with ACT did not differ significantly between high-and low-risk stage II CRC patients [22]. Fu et al. [3] suggested that the value of ACT in stage II colon cancer is much less than previously thought; de-escalating chemotherapy for these patients is necessary. Booth et al. [40] reported that ACT is not related to improved survival for stage II CRC patients with high-risk factors. Although there exist some differences in opinions, we still believe ACT is beneficial for LVI+ patients. The current therapy strategies for N0 stage CRC patients do not directly account for LVI.
To further explain the prognostic value of LVI in CRC patients, we combined LVI with the 8th TNM staging system, which had a better predictive efficacy than the 8th TNM staging system alone. Incorporating the negative impact of LVI into the staging system of CRC may predict the prognosis with greater precision and further establish a more reasonable therapeutic strategy for stage II CRC patients.
This study had some limitations. This was a retrospective study from a single center, and the sample size was not sufficiently large, which may have led to selection bias. A multicenter collaborative study with a large sample size may overcome this issue. In addition, we only focused on the phenomenon and the consequences resulting from LVI; thus, it is necessary for us to explore the genetic mechanism underlying LVI, which may provide novel biomarkers and establish new tumor therapeutic strategies for CRC.

Conclusions
LVI, TB, and PNI are independent prognostic risk factors for stage II CRC patients. Stage II CRC with inadequate ELNs and LVI+ could benefit from adjuvant chemotherapy. The inclusion of LVI can improve the predictive accuracy of the 8th TNM staging system for CRC prognosis. ACT in stage II CRC patients with LVI+ is beneficial for survival. Table 4 Comparison of the performance of the 8th edition of the TNM Staging System alone and the 8th edition of the TNM Staging System combined with LVI. We combined the LVI with the 8th TNM staging system. A comparison was made to estimate the prognostic value between the new system and the 8th TNM staging system. As listed in Table 4, the 8th TNM staging system combined with LVI had a higher C-index than the 8th TNM staging system alone (C-index, 0.895 vs. 0.833), which indicates a better prognostic value for CRC patients